Received: 10 March 2020 | Revised: 28 August 2020 | Accepted: 30 August 2020 DOI: 10.1111/zsc.12450

ORIGINAL ARTICLE

Multilocus and morphological analysis of south-eastern Iberian Wall (, )

Marta Bassitta1 | Joana M. Buades1 | Ana Pérez-Cembranos2 | Valentín Pérez-Mellado2 | Barbara Terrasa1 | Richard P. Brown3 | Pilar Navarro4 | Javier Lluch4 | Jesús Ortega5 | Jose A. Castro1 | Antònia Picornell1 | Cori Ramon1

1Laboratori de Genètica, Departament de Biologia, Universitat de les Illes Balears, Abstract Palma de Mallorca, The phylogenetic relationships among the wall lizards of the 2Departamento de Biología , complex that inhabit the south-east (SE) of the Iberian Peninsula and other lineages of Universidad de Salamanca, Salamanca, the complex remain unclear. In this study, four mitochondrial and two nuclear mark- Spain 3 ers were used to study genetic relationships within this complex. The phylogenetic School of Biological & Environmental Sciences, Liverpool John Moores analyses based on mtDNA gene trees constructed with ML and BI, and a tree University, Liverpool, United Kingdom using *BEAST support three divergent clades in this region: the Valencia, Galera 4 Departamento de Zoología, Facultad and Albacete/Murcia lineages. These three lineages were also corroborated in spe- de Ciencias Biológicas, Universitat de València, Valencia, Spain cies delimitation analyses based on mtDNA using bPTP, mPTP, GMYC, ABGD and 5Evolutionary Ecology Unit, Department of BAPS. Bayesian inference species delimitation method (BPP) based on both nuclear Biology, Lund University, Lund, Sweden data and a combined data set (mtDNA + nuclear) showed high posterior probabilities for these three SE lineages (≥0.94) and another Bayesian analysis (STACEY) based Correspondence Marta Bassitta, Laboratori de Genètica, on combined data set recovered the same three groups in this region. Divergence Departament de Biologia, Universitat de les time dating of the species tree provided an estimated divergence of the Galera line- Illes Balears, Crta. de Valldemossa, km 7.5, 07122, Palma de Mallorca, Illes Balears, age from the other SE group ( , (Albacete/Murcia, Valencia)) at Spain. 12.48 Ma. During this period, the Betic–Rifian arc was isolated, which could have Email: [email protected] caused the isolation of the Galera form distributed to the south of the Betic Corridor.

Funding information Although lizards from the Albacete/Murcia and Galera lineage are morphologically European Regional Development similar, they clearly represent distinct genetic lineages. The noteworthy separation Fund; Consellería d'Educació, Cultura i of the Galera lineage enables us to conclude that this lineage must be considered as Universitats (Govern de les Illes Balears), Grant/Award Number: FPI/1772/2015; a new full species. Ministerio de Economía y Competitividad, Grant/Award Number: CGL-2015- KEYWORDS 68139-C2 Iberian Peninsula, mtDNA, nuclear genes, phylogenetics, Podarcis, South-eastern

1 | INTRODUCTION exception of (Laurenti, 1768), form a mono- phyletic group (Arnold, Arribas, & Carranza, 2007; Carranza, Morphological and genetic studies spanning the last 25 years Arnold, & Amat, 2004) that is generally referred to as the have attempted to clarify relationships both within and be- Podarcis hispanicus (Steindachner, 1870) species complex. tween lineages of the major lacertid genus Podarcis Currently, seven genetically distinct lineages within this from the Iberian Peninsula and North . These studies complex have been raised to species level: Podarcis gua- reported that all Podarcis that inhabit this region, with the darramae, Podarcis virescens, , Podarcis

Zoologica Scripta. 2020;00:1–16. wileyonlinelibrary.com/journal/zsc © 2020 Royal Swedish Academy of Sciences | 1 2 | BASSITTA et al. bocagei, , Podarcis vaucheri and the clustered with those from P. liolepis, in agreement with Pinho nominal taxon P. hispanicus (Caeiro-Dias et al., 2018). The et al. (2006) and unlike the results found by Renoult et al. description of some of these species has been given in differ- (2009) which established that P. liolepis and P. hispanicus ent works, so Harris and Sa-Sousa (2002) found molecular from Galera were two clearly independent groups. The clus- differences between two morphotypes (named types 1 and ter composed by P. liolepis and the Galera lineage was esti- 2) from Western Iberia that were later described as two full mated to have diverged from other members of the species species (Geniez, Sa-Sousa, Guillaume, Cluchier, & Crochet, complex 9.44–13.94 Ma (Kaliontzopoulou et al., 2011). All 2014): Podarcis guadarramae and P. virescens, respectively. these studies trying to establish the phylogeny of Podarcis Moreover, they also divided P. guadarramae into two sub- from the SE (Galera, Valencia and Albacete/Murcia lineages) species: Podarcis guadarramae guadarramae and Podarcis are based on a very low number of individuals for each lin- guadarramae lusitanicus. Harris and Sa-Sousa (2002) also eage. Consequently, the phylogenetic relationships between described a further morphotype (type 3) that was later ele- these groups remain unsolved and require a re-examination vated to species rank as P. liolepis (Renoult, Geniez, Bacquet, based on more extensive information. Guillaume, & Crochet, 2010). Previously, other Iberian For many years, the designation of mitochondrial lineages Podarcis were raised to species level, including P. bocagei within the P. hispanicus complex has been identified solely (Lopez-Seoane, 1885), P. carbonelli (Pérez-Mellado, 1981) by numbers and has coexisted with numerous systematic pro- and P. vaucheri (Busack, Lawson, & Arjo, 2005). posals for these lineages (Geniez et al., 2007, 2014). Despite The nominal taxon within the species complex P. his- the large number of studies carried out on this species com- panicus (Steindachner, 1870) was described as from the plex, the phylogenetic relationships within the complex are south-east (SE) of the Iberian Peninsula (Geniez, Cluchier, not well established and there could be undiscovered inde- Sá-Sousa, Guillaume, & Crochet, 2007). Within this group, pendent lineages. The adequate assignment of potentially Pinho, Ferrand, and Harris (2006) described a new mtDNA new species is a valuable instrument for conservation (Geniez lineage from Galera locality, placed in the Baza Depression et al., 2007, 2014; Renoult et al., 2010). of SE Spain, represented by only a single specimen. This In this paper, we investigate the P. hispanicus complex mtDNA lineage and that of P. liolepis clustered separately with the main aim of establishing the phylogenetic relation- to other members of the species complex. Later work using ships within the SE forms of Podarcis (Valencia, Galera allozyme markers (Pinho, Harris, & Ferrand, 2007) largely and Albacete/Murcia lineage) and between them and other corroborated the existence of a differentiated group in Galera Podarcis from the Iberian Peninsula and North Africa. This area. In these first studies, monophyly of the Galera speci- is achieved through analyses of mtDNA, nuclear DNA and mens with respect to other members of the species complex morphology. These analyses will contribute to knowledge of could not be addressed, due to the inclusion of individuals the evolutionary history and of Iberian Podarcis, from a single location. Later genealogies based on two nu- providing deeper biogeographical insights and important in- clear introns (Pinho, Harris, & Ferrand, 2008) did not support formation for conservation bodies. Podarcis from Galera as a monophyletic group and pointed to nuclear gene flow between the Galera and other lineages identified by mtDNA analyses, where they are in sympatry, 2 | MATERIAL AND METHODS at least across parts of their distributions. Renoult, Geniez, Bacquet, Benoit, and Crochet (2009) identified three evolu- 2.1 | Sampling tionary lineages (P. virescens, P. liolepis and P. hispanicus) in the east of Iberian Peninsula from morphological char- In total, 105 Podarcis individuals from the Iberian Peninsula acters and nuclear loci, while their analysis of mtDNA re- and North African region were captured (Table S1 and vealed four lineages (P. virescens, P. liolepis, P. hispanicus Figure 1). Lizards were caught by careful noosing in their from Galera and P. hispanicus from Valencia) suggesting natural with the specific permits delivered by the an ancient introgression. Other studies added to the Galera competent body in each locality. All morphological measure- (Pinho et al., 2006) and Valencia (Renoult et al., 2009) forms, ments were taken in situ, and 1 cm of the tail tip was removed an additional mitochondrial lineage detected in SE Spain and stored in 100% ethanol. All lizards were released at their from the Albacete/Murcia area. The Valencia and Albacete/ capture site. Murcia populations appear to comprise sister mitochondrial Because the distribution of different lineages of P. hispan- lineages, which together represent a sister group to all North icus complex is not well-known in the SE Iberian Peninsula African lineages, from which they diverged 6.99—9.44 Ma (see, for example, Caeiro-Dias et al., 2018 and references (Kaliontzopoulou, Pinho, Harris, & Carretero, 2011). In therein), we sampled some of the localities used by previ- that study, a second individual from the Galera lineage was ous authors, where particular lineages were detected and de- included and the results indicated that this mtDNA lineage scribed. In addition, we tried to include new localities within BASSITTA et al. | 3

a Grossa 59-65 99 94 Mancolibre 92,93 96 73 70-71 95 100-103 74 72 7 4,5 97 104,105 COLUMBRETES ISLANDS

12,24 Lobo 98 Foradada 11 68,69 66,67 8,14 15-18 29-32 b 6 3 9 10 19-23 48,49 75-90 91 a 25-28 46,47 45 b 24 52,53 50,51 54 56-58 55 36,37 41,42 38-40 43,44 CUENCA 48,49 33-35

VALENCIA 46,47 45

52,53 1 ALBACETE 50,51 54 ALICANTE

Laderas del Campillo 55 41,42 38-40 56-58 36,37 Monteagudo Galera MURCIA 43,44 33-35

GRANADA

2

P. muralis P. bocagei P. carbonelli P. guadarramae lusitanicusP. guadarramae guadarramae P. virescens P. liolepis P. vaucheri GaleraAlbacete/Murcia Valencia

FIGURE 1 Locations of all Podarcis samples from the Iberian Peninsula and North Africa used in this study. The colours correspond to the different species indicated in the legend below and the numbers correspond to those in Table S1. The black triangle marks the locality type of Podarcis hispanicus sensu stricto (Monteagudo, Murcia) the expected range of each lineage. Similar to previous au- four non-overlapping mtDNA fragments were obtained for thors (Geniez et al., 2007), we intensively searched the re- each specimen: (a) partial 12S rRNA (360 bp), (b) partial stricted type locality of P. hispanicus (Steindachner, 1870), control region (CR) (476 bp), (c) two partial fragments of Monteagudo, close to Murcia town, twice in 2019. Despite cytochrome b (CYTB) (306 and 483 bp, respectively) and suitable climatic conditions, we did not find any Podarcis at (iv) two partial subunits of the NADH dehydrogenase gene the type locality but did detect a small population at Laderas and associated tRNAs (referred to as ND1 (48 bp), ND2 del Campillo, 4.5 km north from Monteagudo (Figure 1). To (415 bp), tRNAIle, tRNAGln and tRNAMet (213 bp)). Two our knowledge, this is the closest location to the type locality of partial nuclear genes were amplified and sequenced: (a) P. hispanicus sensu stricto (Geniez et al., 2007). Lizards from melanocortin-1 receptor gene (MC1R) (663 bp) and (b) re- Galera (Granada) were captured with a Permit of Scientific combination-activating gene 1 (RAG1) (939 bp) (Table S1). Capture from the Consellería de Agricultura, Ganadería, Primers and amplification conditions are the same as those Pesca y Desarrollo Sostenible of Junta de Andalucía (permit used in our previous studies of Podarcis (Buades et al., 2013; No: 201999900530233 issued on 24/07/2019). Rodríguez et al., 2013, 2014, 2017; Terrasa et al., 2009). Both strands of the PCR products were sequenced on an au- tomated ABI 3130 sequencer (Applied Biosystems) using a 2.2 | DNA extraction, BigDye® Terminator v. 3.1 Cycle Sequencing Kit (Applied amplification and sequencing Biosystems) and edited using CodonCode Aligner software (CodonCode Corporation). Nuclear data were phased using A standard phenol–chloroform protocol was used for the PHASE algorithm (Stephens, Smith, & Donnelly, 2001) DNA extraction (González et al., 1996). The following within DnaSP v.6 (Librado & Rozas, 2009). Eighty-five 4 | BASSITTA et al. sequences that had been published by previous studies were 2.4 | Species tree and divergence also used (Table S1). times estimates

A standard likelihood ratio test of the molecular clock was 2.3 | Phylogenetic analyses performed for both mtDNA and nuclear data sets in MEGA7 (Kumar, Stecher, & Tamura, 2016). This test reliably informs We identified 88 unique haplotypes within the concatenated whether a strict or relaxed clock model is most suitable for mitochondrial DNA data set (2,301 bp) from the Podarcis spec- divergence time dating (Brown & Yang, 2011). imens with DnaSP v.6 (Librado & Rozas, 2009). Individuals The species tree approach that is implemented in *BEAST from the Balearic Islands were included as out-groups (five (Heled & Drummond, 2010) was used in an attempt to simul- Podarcis lilfordi and three Podarcis pityusensis). Sequences taneously infer the phylogenetic relationships and divergence were aligned in the MAFFT v7.423 online server (Katoh times between the different lineages of the Iberian and North & Toh, 2008) using the iterative refinement method (FFT- African Podarcis based on: (a) all mtDNA sequence data NS-i). Best-fit nucleotide substitution models and partition- sets (108 individuals) and (b) both mtDNA and the phased ing scheme were chosen simultaneously using PartitionFinder nuclear loci (mtDNA + nuclear) (RAG1: 95 individuals and v2.1.1 (Lanfear, Frandsen, Wright, Senfeld, & Calcott, 2016) MC1R: 102 individuals). For mtDNA analyses, the species under the Akaike information criterion (AIC). The partitioning tree was calibrated using a mean split time of 5.3250 Ma schemes were defined manually (by gene and by codon), with with relatively little uncertainty around this estimate, rep- branch lengths of alternative partitions ‘unlinked’ to search for licating the calibration in Rodríguez et al. (2013). The cal- the best-fit scheme. The proportion of invariable sites (I) pa- ibration was specified from a log-normal distribution with rameter was discarded if the favoured model incorporated both mean 1.6724 and standard deviation 0.002 (the central 95% the I and the Gamma site rate heterogeneity (G) parameters as of this distribution ranges from 5.304 to 5.346, in real space). simultaneous use of these parameters can have undesirable ef- This calibration is based on knowledge of the timing of the fects (Yang, 1993). end of the Messinian salinity crisis (5.33 Ma) and the very We performed phylogenetic analyses using maxi- rapid refilling of the Mediterranean basin that would have mum-likelihood (ML) and Bayesian inference (BI) meth- separated the two Balearic island Podarcis (i.e. P. lilfordi and ods. Maximum-likelihood analyses were performed using P. pityusensis; see Brown et al., 2008 and references therein). IQ-TREE version 1.6.10 (Nguyen, Schmidt, von Haeseler, Three partitions were assigned as: (a) 12S rRNA, CR, all & Minh, 2014). We applied the partitions and the best-fit tRNAs (b) CYTB/ND1/ND2 1st + 2nd codon position and substitution model and performed 106 bootstrap replicates (c) CYTB/ND1/ND2 3rd codon position. Evolutionary mod- based on the ultrafast bootstrap approximation (UFBoot) els were the same as those used for MrBayes. The *BEAST (Hoang, Chernomor, Von Haeseler, Minh, & Vinh, 2017; MCMC sampler was run twice for 5 × 108 generations, with Minh, Nguyen, & von Haeseler, 2013) for statistical one step per 50,000 being sampled. A relaxed log-normal support. clock model was specified since the molecular clock likeli- Bayesian analyses were performed with MrBayes 3.2.6 hood ratio test indicated rate variation across the gene trees. (Ronquist et al., 2012). MCMC chain lengths were 107 gen- A coalescent Yule speciation process was used for the tree erations with a sampling frequency of 103 generations. We prior. For mtDNA + nuclear analyses, we used the same cal- used two simultaneous runs of three hot and one cold chain ibration described for the mtDNA-only analyses. The same each. Convergence was confirmed by examining the station- DNA substitution model was used for the three mtDNA par- arity of the log-likelihood (lnL) values of the sampled trees titions and the JC69 model was used for the two nuclear loci. and the observation of average standard deviations of the split Tracer v1.7 (Rambaut et al., 2018) was used to check for con- frequencies being <0.01. Run characteristics such as effec- vergence. Posterior trees were combined to obtain the tree tive sample sizes (ESS) were also assessed in Tracer v1.7 with the maximum sum of posterior clade probabilities using (Rambaut, Drummond, Xie, Baele, & Suchard, 2018). The mean heights for node annotation. fifty per cent majority-rule consensus tree was summarized using sumt command with the first 25% of each run discarded as burn-in. The resulting phylogenetic tree was visualized 2.5 | Genetic structure and species and edited using FigTree v1.4.2 (Rambaut, 2014). delimitation analyses Haplotype networks were constructed for each phased nuclear locus using the Population Analysis with Reticulate Two tree-based (bPTP, mPTP and GMYC) and one distance- Trees (PopART, http://popart.otago.ac.nz) (Leigh & Bryant, based (ABGD) species delimitation methods were performed 2015) with the TCS method (Clement, Posada, & Crandall, on the concatenated mitochondrial haplotypes data set. bPTP 2000). (Zhang, Kapli, Pavlidis, & Stamatakis, 2013) analyses were BASSITTA et al. | 5 0 0 0 0 0.20 0.41 0.09 0.23 0 0 0 0 0 0 P [10] 0.07 0.17 0.05 0.15 0.62 0.41 0.57 0.52 0.05 0.09 0.11 0.18 0.05 0.09 P [11] 0.93 0.82 0.95 0.85 0.18 0.11 0.34 0.22 0.95 0.91 0.89 0.82 0.95 0.91 P [12] PP for number of species 0.01 0.09 0.01 0.05 0.18 0.26 0.18 0.24 0 0.05 0 0.10 0.00 0.04 P [Vir- Car] 0.06 0.08 0.04 0.07 0.48 0.22 0.37 0.27 0.05 0.03 0.11 0.06 0.04 0.03 P [Boc-Car] 0.99 1.00 1.00 1.00 0.88 0.69 0.92 0.85 1.00 1.00 1.00 1.00 1.00 1.00 P [Vau] 0.93 0.84 0.95 0.86 0.23 0.20 0.38 0.31 0.95 0.91 0.89 0.83 0.95 0.92 P [Car] 0.94 0.92 0.96 0.93 0.44 0.60 0.59 0.62 0.95 0.97 0.89 0.94 0.96 0.97 P [Boc] 0.99 0.92 0.99 0.93 0.70 0.45 0.77 0.61 1.00 0.95 1.00 0.90 1.00 0.95 P [Vir] 0.99 0.99 1.00 1.00 0.76 0.54 0.87 0.71 1.00 1.00 1.00 0.99 1.00 1.00 P [Lus] 1.00 0.98 1.00 0.97 0.98 0.94 0.99 0.96 1.00 1.00 1.00 1.00 1.00 1.00 P [Val] 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 P [Gua] 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 P [Mur] 1.00 1.00 1.00 1.00 1.00 0.98 1.00 0.99 1.00 1.00 1.00 1.00 1.00 1.00 P [Alb] 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 P [Lio] 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 P [Gal] 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 PP of each species P [Bal] mt + nuclear Nuclear mt + nuclear Nuclear mt + nuclear Nuclear mt + nuclear Nuclear mt + nuclear Nuclear mt + nuclear Nuclear mt + nuclear Nuclear Results from BPP analyses for each combination of priors based on phased nuclear loci only (first row) and combined data set mtDNA + nuclear (second G (3. 0.04) G (3. 0.04) G (3. 0.004) G (3. 0.004) G (3. 0.2) G (3. 0.2) G (3.0.02) G (3.0.02) G (3.0.02) G (3.0.02) G (3. 0.2) G (3. 0.2) G (3. 0.004) G (3. 0.004) τ 0 0.002) 0.002) 0.002) 0.002) G (3. G (3. G (3. G (3. G (3. 0.2) G (3. 0.2) G (3. 0.2) G (3. 0.2) G (3. 0.02) G (3. 0.02) G (3. 0.02) G (3. 0.02) G (3. 0.02) G (3. 0.02) Priors θ TABLE 1 Note: The posterior probability (PP) values are averages from three independent runs of module A11. Abbreviations: Bal, Balearic outgroup; Car, Podarcis carbonelli ; Gal, Galera lineage; Lio, liolepis Alb, Albacete/Murcia Lus, guadarramae lusitanicus Mur, muralis Gua, guadarramae ; Val, Valencia lineage; Vau, Podarcis vaucheri Vir, virescens Boc, bocagei . 6 | BASSITTA et al. performed using the online server (https://speci​es.h-its.org/) a robust means of inferring species from recently diverged and the ML tree from IQ-TREE. We ran 500,000 generations lineages using multilocus data (Leaché, Zhu, Rannala, & with a thinning of 500 and a burn-in of 0.1, then assessed Yang, 2018; Rannala, 2015) under the multispecies coales- convergence visually using the MCMC iteration vs. log- cent model (MSC) (Rannala & Yang, 2003). This Bayesian likelihood plot automatically generated. Next, we applied the method assumes no gene flow among species and no recom- mPTP method (Kapli et al., 2017) using the ML tree from bination among loci and explains gene tree discordance by in- IQ-TREE. We performed two simultaneous Markov chain complete lineage sorting (ILS). BPP implements a reversible Monte Carlo (MCMC) runs of 100,000,000 steps, sampling jump Markov chain Monte Carlo (rjMCMC) search to esti- every 10,000 steps. Convergence was confirmed with the mate the sum of posterior probability (PP) of all species de- likelihood plot of the combined runs and the average standard limitation models considered. The speciation models PP can deviation of delimitation support values (0.000009), which in- be affected by the prior distributions chosen for the ancestral dicate convergence on the same distribution as it approaches population size (θ) and root age (τ). We used a range of dif- to zero. The -multi option was used to allow differences in ferent prior scenarios (Table 1), including small population rates of coalescence among species. Finally, we incorporated size (θ) and deep divergence (τ), or large population size and the single-threshold GMYC (Fujisawa & Barraclough, 2013; shallow divergence that favours more conservative models Pons et al., 2006) model. This model is based on an ultramet- containing fewer species (Yang & Rannala, 2010). We used ric phylogenetic tree and uses differences in branching rates the A11 model (speciesdelimitation = 1 and speciestree = 1), to infer species delimitation. The input tree was generated in which jointly infers the species tree and the species delimita- BEAST v2.6 (Bouckaert et al., 2019) under a relaxed clock tion (Rannala & Yang, 2017). The assignment of individuals and Yule model tree prior. The same partitioning model used to hypothesized species was based on the identified mtDNA for the ML and BI trees was used as estimated using AIC in lineages. The MCMC chain was run for 104 steps (following PartitionFinder v2.1.1 (Lanfear et al., 2016). Analyses were a burn-in of 2,000 steps), sampled every 25 steps. Each anal- run for 200 million generations, sampling every 2,000 gen- ysis was run three times to confirm that they converged on erations. Convergence and mixing were monitored in Tracer the same posterior. v1.7 (Rambaut et al., 2018), and ESS values >200 indicated Finally, we used STACEY v1.8.0 (Jones, 2017), a adequate sampling of the posterior. TreeAnnotator within Bayesian multispecies coalescent method implemented in BEAST was used to create a tree with the maximum sum BEAST v2.5 (Bouckaert et al., 2019) that allows inference of posterior clade probabilities using mean heights for node of species delimitation and species phylogeny. The author of annotation and a 10% of burn-in. The GMYC analysis was this analysis prefers the term ‘cluster’ over ‘species’ (Jones, conducted using the single-threshold method in the splits 2017), and so, we use both terms when discussing this analy- package on R v. 3.6.1. sis. STACEY provided a parallel analysis to that in BPP and Automatic Barcode Gap Discovery (ABGD; Puillandre, so was particularly useful to ensure the robustness of the spe- Lambert, Brouillet, & Achaz, 2012) uses pairwise genetic cies delimitation results. STACEY was preferred over a sim- distances to determine the differences between intra- and in- ilar approach within *BEAST (Grummer, Bryson, & Reeder, terspecific divergence and establishes the number of different 2014; Heled & Drummond, 2010) because it provided bet- species based on those differences. ABGD was performed ter convergence of the MCMC chain. It differs from BPP in using the online tool (http://wwwabi.snv.jussieu.fr/publi​ ​c/ several ways, including the implementation of a birth–death– abgd/abgdweb) with default values for the prior intraspecific collapse model, which eliminates the need for a rjMCMC divergences (Pmin 0.001, Pmax 0.1, steps 10), relative gap algorithm, and use of different population size parameters width (1.5) and distance distribution (20). Results were com- which are integrated out. As for BPP analyses, we used the pared using both JC69 and K80 models. three loci comprised mtDNA and the two-phased nuclear loci Structural analysis was performed with the Bayesian with the JC69 substitution model and a strict clock specified model-based clustering algorithm implemented in BAPS for each. Two independent runs of 450 million generations 6.0 (Corander & Tang, 2007) using the mtDNA haplotypes of the MCMC chains were performed sampling every 5,000 alignment. The method was performed using clustering with steps. Convergence was checked in Tracer 1.7 (Rambaut linked loci and codon linkage model. We ran BAPS for values et al., 2018), and posterior trees from both runs were com- of k ranging from one to fifteen, performing three replicates bined (the first 20% of trees from each run were removed for each value of k. as burn-in). The resulting 144,000 species trees were pro- Bayesian Phylogenetics and Phylogeography (BPP cessed in SpeciesDelimitationAnalyser (Jones, 2017) with a v4.0) (Yang & Rannala, 2010) analysis was performed collapse height of 0.0001 (the same value used in STACEY on two different data sets: (a) two-phased nuclear loci and analysis) and default similarity cut-off (0.9). Posterior trees (b) mtDNA + two-phased nuclear loci. We also included were also combined to obtain a maximum sum of clade cred- Balearic Podarcis in the analysis as the out-group. BPP is ibilities tree. BASSITTA et al. | 7 2.6 | Morphological analyses trees clustered the individuals into four major mitochon- drial clades within the Iberian Peninsula and North African To avoid any confusion regarding their lineage assignation, samples, corresponding to the following taxa/areas: group 1 our morphometric analyses were restricted to individuals that (monophyletic) only includes P. muralis; the other groups are we personally generated sequence data for. Thus, we ana- polyphyletic including group 2: P. carbonelli, P. virescens, lysed a rather small sample of lizards (25 individuals from P. g. lusitanicus, P. bocagei and P. g. guadarramae; group 3: Galera lineage and 12 belonging to Albacete/Murcia line- P. vaucheri, Galera lineage, Valencia lineage and Albacete/ age). Our aim was only to show general similarities and dif- Murcia lineage; and group 4: P. liolepis (Figure 2). The sepa- ferences between the two lineages of Albacete/Murcia and ration between groups 3 and 4 presents low support (poste- Galera. A more thorough analysis will be subsequently car- rior probability < 50; bootstrap support < 40). Divergence ried out with a larger sample of individuals from the entire of the different SE mitochondrial lineages (Galera, Albacete/ Iberian Podarcis species complex. Murcia and Valencia) show high support (PP ≥ 0.99; Six body dimensions, as well as body mass (Weight), BS ≥ 79). The results show Valencia and Albacete/Murcia as were included in this study: snout–vent length (SVL), in- sister lineages to the North African form (P. vaucheri), while tact tail length (TL), pileus length (PL), head width (HW), the entire group (P. vaucheri, (Albacete/Murcia, Valencia)) head height (HH) and left hindlimb length (HLL). All mea- is a sister lineage to the Galera lineage. surements were made with a digital calliper to the nearest Networks based on phased nuclear loci (Figure 3) show 0.01 mm, except for SVL and intact tail length, which was less structured interrelations between different Podarcis pop- measured with a steel rule to the nearest 1 mm. Weight ulations. Podarcis muralis and P. vaucheri are the only groups was obtained with a spring scale Pesola®. Six scalation that share no haplotypes with any other Podarcis populations. characters were recorded: gularia, collaria, dorsalia, ven- The Galera lineage presents two shared haplotypes and nine tralia, femoralia and left fourth digit lamellae (see Pérez- species-specific haplotypes in the RAG1 network, and one Mellado & Gosá, 1988) for methodological details of body shared haplotype and five unique haplotypes in the MC1R measurements and scalation counts). Not all characters network. These haplotypes were shared with geographically could be recorded from all individuals. Due to sexual di- close groups (P. virescens, P. liolepis, Valencia and Albacete/ morphism, males and females were analysed separately. Murcia lineages), except for P. g. lusitanicus, indicating a Raw or log-transformed data were checked for normality common haplotype among these populations with posterior (Shapiro–Wilk test) and homogeneity of variances (Fligner differentiation. The striking number of different haplotypes test) prior to statistical comparisons (one-way ANOVA). If found in P. liolepis demonstrates the high genetic diversity assumptions for the use of parametric techniques were not present in this group and could be an indicator that its distri- met, we employed non-parametric equivalents. All analy- bution range has not been fully studied genetically. ses were done within R (R Core Team, 2018). In addition, we studied colour and design variation in our samples, em- ploying the same criteria as Geniez et al. (2007) in their 3.2 | Species tree and divergence diagnostic descriptions. Colour descriptions were com- time estimates pleted with the assignment of colour codes to the follow- ing body parts: background dorsum, pileus, flanks, belly, The mtDNA species tree (Figure 4) provided a posterior gular region, blue ocelli in outer ventral scales, if present, mean for the divergence of the Iberian and North African dorsolateral stripes, dorsal side of the tail and ventral side group with respect to the monophyletic group P. muralis of the tail. Colour codes were assigned according with the at 15.60 Ma (95% highest posterior density, HPD: 22.31– catalogue of Köhler (2012). 10.71 Ma) and that for the divergence of Galera from the clade containing (P. vaucheri (Albacete/Murcia, Valencia)) at 12.48 Ma (95% HPD: 18.27–7.93 Ma). The Valencia and 3 | RESULTS Albacete/Murcia lineage divergence appeared to be more re- cent at 7.11 Ma (95% HPD: 11.75–1.87 Ma). The split be- 3.1 | Phylogenetic analysis tween P. liolepis and the remaining lineages from west Iberia would have taken place during the middle Miocene, in the PartitionFinder v2.1.1 (Lanfear et al., 2016) identified three late Serravallian (12.58 Ma, 95% HPD: 18.53–7.40 Ma), partitions with the following substitution models: non-coding as also observed for the divergence of the Galera lineage. fragments [GTR+I+G], 1st and 2nd codon of coding regions Despite the use of simpler models with fewer parameters, [HKY+I+G] and 3rd codon of coding regions [GTR+I+G] the mtDNA + nuclear analyses did not show repeatable con- (analyses were carried out with these models but without vergence on the same posterior, and thus, the results are not the I parameter, as discussed in the section 2). ML and BI presented here. 8 | BASSITTA et al.

ABGD mPTP bPTP GMYC BAPS AB7_P. liolepis AM2_P. liolepis AB5_P. liolepis AB2_P. liolepis AB1_P. liolepis AB0_P. liolepis AF2_P. liolepis PL60_P. liolepis HJ4_P. liolepis MPC5_P. liolepis HP10_P. liolepis MPC6_P. liolepis HP8_P. liolepis HP218_P. liolepis HP9_P. liolepis Group 4 HPC8_P. liolepis HJ3_P. liolepis HJ2_P. liolepis HP118_P. liolepis 100 HP4_P. liolepis 1.00 HP3_P. liolepis 86 HP5_P. liolepis 0.65 HP11_P. liolepis 100 HF9_P. liolepis 1.00 PL3_P. liolepis HL1_P. liolepis Alb3_Albacete/Murcia Alb6_Albacete/Murcia Alb1_Albacete/Murcia 100 Alb4_Albacete/Murcia Seg2_Albacete/Murcia 1.00 Cmp1_Albacete/Murcia * 97 PHP3_Valencia 0.51 0.99 PHP4_Valencia 100 100 PHCE2_Valencia PHCE4_Valencia 1.00 1.00 HB2_Valencia 100 Vau01_P.vaucheri Vau02_P.vaucheri 1.00 Group 3 Ph22_Galera 79 Ro2_Galera Ro1_Galera 0.99 Ph16_Galera AB9_Galera Ph21_Galera AB8_Galera 100 Df3_Galera 1.00 Gale1_Galera 94 Df1_Galera HGF_P. g. guadarramae 0.99 HGG5_P. g. guadarramae HGF2_P. g. guadarramae HGG4_P. g. guadarramae HGG1_P. g. guadarramae HT1_P. g. guadarramae 100 HGF5_P. g. guadarramae 1.00 HGG2_P. g. guadarramae HGF1_P. g. guadarramae HE4_P. g. guadarramae 100 HM3_P. g. guadarramae 1.00 HS1_P. g. guadarramae * HC1_P. g. guadarramae 99 0.76 100 BN3_P. bocagei Group 2 1.00 BN5_P. bocagei 1.00 100 CAR02_P. g. lusitanicus 100 HN1_ 100 1.00 P. g. lusitanicus 1.00 HVA1_P. virescens 1.00 100 HVA2_P. virescens 100 1.00 HVP3_P. virescens 100 1.00 HVP2_P. virescens HD1_P. virescens 1.00 CAR01_P. carbonelli MPG4_P. muralis MSS1_P. muralis MSP2_P. muralis MVB1_P. muralis 100 MPN1_P. muralis MSP1_P. muralis 90 0.99 MGV1_P. muralis 0.99 MLL1_P. muralis 100 Mpfn7_P. muralis 100 1.00 MLP1_P. muralis Group 1 MP2_P. muralis 1.00 MP11_P. muralis 99 MP7_P. muralis MP14_P. muralis 1.00 MP3_P. muralis MP13_P. muralis Cp1_P. lilfordi 100 100 SCX3_P. lilfordi 100 1.00 Cr1_P. lilfordi 1.00 D22_P. lilfordi 100 1.00 AG1_P. lilfordi Alg1_P. pityusensis

1.00 Outgroup 100 Fort4_P. pityusensis 1.00 Eiv1_P. pityusensis

FIGURE 2 Gene tree based on maximum likelihood for mitochondrial data showing different scenarios of species delimitation between Iberian and North African Podarcis populations, using the Balearic clade as an out-group. The numbers above branches correspond to bootstrap support and numbers below branches are posterior probabilities from Bayesian analysis. Support values for nodes with ≤50% support are not included. Asterisks indicate branches where different analyses produced different topology. Group 1: Podarcis muralis; Group 2: Podarcis bocagei, Podarcis carbonelli, Podarcis guadarramae guadarramae, Podarcis guadarramae lusitanicus, Podarcis virescens; Group 3: Podarcis vaucheri, Galera lineage, Albacete/Murcia lineage, Valencia lineage; and Group 4: Podarcis liolepis 3.3 | Genetic structure and species and a deep/moderate divergence time (τ0 InvG (3, 0.2)) or delimitation analysis (τ0 InvG (3, 0.02)), which provided the greatest support for 11 species (0.520 and 0.62, respectively). Podarcis bo- The different species delimitation methods (bPTP, mPTP, cagei and P. carbonelli are the two populations that are not GMYC, ABGD and BAPS), based on mtDNA, returned dif- clearly defined as separate species (Table 1 and Figure 2). ferent partition numbers, ranging from 9 to 22. All the meth- In terms of species delimitation, all prior combinations ods corroborated the separation of the SE lineages (Valencia, showed strong individual species support (≥0.94) for the Albacete/Murcia and Galera; Figure 2). Galera, Albacete/Murcia and Valencia lineages (Table 1). Under five out of seven prior specifications for θ and τ, Nonetheless, the inferred species tree topologies tended to BPP analyses for both data sets nuclear and combined vary among models and even (to a lesser extent) showed mtDNA + nuclear loci provided the greatest support for some minor differences between runs for a given prior the delimitation of 12 species (Table 1). The analyses in combination. Posterior support for nodes within the BPP which the parameters favoured a moderate populations size species trees based on both nuclear data and mtDNA + nu- (θ: InvG (3, 0.02)) and divergence time (τ0: InvG (3, 0.02)) clear combined data was generally low, except for P. mu- provided the highest posterior probability (0.91 and 0.95, ralis and P. vaucheri, and the topology differed from that respectively) for 12 species. This contrasts with analyses described by phylogenetic trees based on the mtDNA con- that favoured a larger population size (θ: InvG (3, 0.2)) catenated data set.

FIGURE 3 TCS haplotype networks of Iberian and North African Podarcis nuclear loci for melanocortin receptor 1 (MC1R) and recombination-activating gene 1 (RAG1). Hatch marks between black dots represent one mutational step, haplotypes circle area is proportional to the number of individuals and the colour identifies the species BASSITTA et al. | 9

MC1R

RAG1

P. g. guadarramae P. g. lusitanicus P. bocagei P. carbonelli P. virescens P. liolepis Galera Albacete/Murcia Valencia P. vaucheri P. muralis Outgroup 10 | BASSITTA et al. Albacete/Murcia 0.92 [1.87-11.75]

0.88 [5.85-15.38] Valencia

0.79 [7.93-18.27] P. vaucheri

Galera

P. bocagei 0.70 [0.92-8.31] 0.68 [9.65-20.43] 0.90 [3.96-10.50] P. g. lusitanicus

P. g. guadarramae 0.96 [5.32-13.09]

P. carbonelli 0.94 [0.11-8.58] 0.71 [7.40-18.53] 1.00 [10.71-22.31] P. virescens

P. liolepis

P. lilfordi 1.00 [5.31-5.35]

0.88[6.21-18.32] P. pityusensis

P. muralis

MIOCENE PLIOCENE PLEISTOCENE

20 15 10 5 0

FIGURE 4 Species tree with estimated divergence time based on mitochondrial data (2,301 bp of 12S, CR, CYTB and NADH fragments) resolved from *BEAST analyses. The divergence times were estimated using the separation of the Balearic Podarcis clade as a calibration. Horizontal bars and numbers at the right of nodes (Ma) indicate a 95% credible interval of divergence and posterior probability values are shown to the left

Species delimitation analysis performed with STACEY F1,22 = 1.816, p = .192, adult females: F1,11 = 0.173, provided similar results to those obtained using BPP and p = .686) as is body mass (males: F1,19 = 0.0128, p = .991, attributed highest posterior probabilities to 13 clusters females: F1,10 = 3.703, p = .083). The number of lizards (0.96). In this analysis, an extra group was found corre- with an intact tail precluded any comparison between line- sponding to the separation of the out-group into the two ages (Table S2). For head measurements, only head width recognized species from the Balearic islands (P. lilfordi and showed significant differences between males of Albacete/ P. pityusensis). Nonetheless, the three SE lineages from the Murcia and Galera (F1,22 = 9.541, p = .0054) but not be- Iberian Peninsula are confirmed to represent independently tween females (F1,11 = 2.041, p = .181). The remaining evolving lineages. Also similar to BPP analyses, support head measurements were similar in both lineages (pileus values were low for individual internal nodes within the length, males: F1,22 = 1.658, p = .211, females: F1,10 = 0.06, species tree, except for the divergence of P. vaucheri from p = .811; head height, males: F1,22 = 1.832, p = .19, fe- the rest of species complex. males: F1,11 = 1.53, p = .242). Hindlimb length was also similar in both lineages (males: F1,22 = 3.1, p = .0922, fe- males: W = 6, p = .106). 3.4 | Morphological analysis With regard to scalation traits, the number of subdigital lamellae under the 4th toe is similar in both lineages (males: In the SE Iberian Peninsula, both groups of lizards from F1,21 = 0.43, p = .519, females: F1,7 = 1.485, p = .262). the Albacete/Murcia and Galera lineages are characterized There is a greater number of femoral pores in males from by a very small body size (SVL and Weight) in comparison the Albacete/Murcia lineage (Table S2, F1,22 = 16.48, with the remaining recognized species of Iberian Podarcis p = .00052) but not in females (F1,10 = 0.561, p = .471). Also, (Kaliontzopoulou, Carretero, & Llorente, 2012). In both the number of gular scales is greater in Albacete/Murcia −5 lineages, we found an SVL under 53 mm (Table S2). Body males (Table S2, F1,22 = 27.02, p = 3.26 × 10 ) but not in size is statistically similar in the Albacete/Murcia and females (F1,10 = 0.944, p = .354). We did not find differences Galera lizards (Table S2, adult males: one-way ANOVA, in the number of dorsal scales (males: F1,21 = 0.19, p = .668, BASSITTA et al. | 11 females: F1,9 = 2.065, p = .185) nor in collaria (males: arrangement of different lineages in the phylogeny of Iberian F1,22 = 0.132, p = .72, females: F1,9 = 1.205, p = .301). Podarcis (Table 1 and Figures 2 and 4), we can conclude that Finally, the number of ventral scales is significantly higher the Galera lineage represents a new full species: Podarcis in Galera lizards (Table S2, males: Mann–Whitney test, galerai sp. nov. The taxonomic arrangement of this lineage W = 23.5, p = .0123, females: F1,9 = 7.333, p = .024). within the Iberian Podarcis complex and the description of Lizards from both lineages of Albacete/Murcia and the new species in comparison with the nominal taxon P. his- Galera are small dorsally brownish or greyish lizards. The panicus is proposed below. general appearance of the lizards from both lineages is very similar (Figures S1 and S2). In the Galera lineage, we ob- served dorsal colours from Drab (code 19 of Köhler, 2012) 4 | TAXONOMY to Light Drab (269) and, especially, Olive-Brown (278) and Brownish Olive (276) (Table S4). The dorsal background of Family lizards from the Albacete/Murcia lineage is mainly Drab (19), Genus Podarcis Wagler, 1830 Ground Cinnamon (270) or Light Drab (269). Thus, in the Albacete/Murcia lineage we can observe olive nuances that, apparently, are absent in the Galera lineage. A similar pat- 4.1 | Podarcis hispanicus Steindachner, 1870 tern was observed in pileus colour (Table S4). In the Galera lineage, bellies are mainly Smoky White (261) or Light Buff oxycephala var. hispanica Steindachner, 1870. (2), while in the Albacete/Murcia lineage bellies are Pale Sitzungsber. Akad. Wiss. Wien, Math. Naturwiss. Kl., Abt.1, Buff (1) or Light Buff (2) (Table S4). 62(8): 350. Taking into account the same diagnostic characters em- Lacerta muralis steindachneri Bedriaga, 1886. Abh. ployed by Geniez et al. (2007) to define the nominotypical Senckenberg. Naturforsch. Ges., 14(2): 256. taxon, P. hispanicus, we can see that almost all diagnos- Podarcis [hispanicus] hispanicus Steindachner, 1870, tic features of colour and design can be found in both lin- Geniez et al., 2007. Herpet. J., 17:74. eages. However, all these traits are better represented by the Albacete/Murcia lineage (Table S3 and Figure S2). A wider sampling of its geographic region is required to pro- 4.1.1 | Type locality vide an in-depth morphological description of the Valencia lineage. Monteagudo (not Monte Agudo), province of Murcia. Terra typica restricta Geniez et al. (2007). Geniez et al. (2007) designed as the lectotype the spec- 3.5 | Taxonomic results imen NMW 16088:1 (Naturhistorisches Museum Wien, Austria). Geniez et al. (2007) carried out a morphological analy- sis of several lizards from the SE Iberian Peninsula, before the discovery of two new lineages: Albacete/Murcia and 4.1.2 | Diagnosis Valencia (Caeiro-Dias et al., 2018 and references therein). Consequently, it is possible that their sample contained liz- Due to the coincidence of several morphometric traits of the ards from different lineages. We do not know whether this lineages from Galera and Albacete/Murcia, the diagnosis of was the case, but from our results, lizards from the Albacete/ Geniez et al. (2007) is, in general, valid for both lineages, Murcia lineage are closer to the nominotypical taxon, as de- here recognized as separated species. A small lizard with a fined by Geniez et al. (2007) (see Table S3). In addition, at very flattened head (Table S2). Generally, it lacks the mas- the closest locality to Monteagudo, Laderas del Campillo, seteric plate. Dorsum brown. In most lizards, there is a con- we found lizards assigned to the Albacete/Murcia lineage. tinuation of light dorsolateral stripes and dark stripes along Thus, we propose that the Albacete/Murcia lineage could parietal scales of the head. Normally (more than 83% of indi- be considered the present-day representation of the nomino- viduals), vertebral line is present and in the majority of cases, typical taxon, P. hispanicus sensu stricto. In fact, the phy- bi or even trifurcated (Table S3 and Figure S2). Belly with logenetic tree (Figure 2) shows that there is a monophyletic white colour, even if more than 30% of individuals can show group formed by the Valencia lineage and P. hispanicus a yellowish colour, in some cases covering the base of the sensu stricto (Albacete/Murcia lineage). This clade is the tail (Figure S2). In 25% of individuals, blue ocelli are present sister group to P. vaucheri. The Galera lineage is a mono- in outer ventral scales. From 56 to 73 dorsal scales (Geniez phyletic group, clearly separated from the other Eastern et al., 2007, mention lizards with a minimum of 44 dorsal lineages of Iberian Podarcis. In line with this result and the scales). 12 | BASSITTA et al. 4.2 | Podarcis galerai sp.nov 4.2.5 | Diagnosis 4.2.1 | Holotype A small wall lizard with <53 mm of SVL (see Table S2 for averages, maximum and minimum values of morpho- Herpetological collection (Colección Herpetológica de la metric and scalations characters). Dorsal pattern and col- Universidad de Salamanca, CHUS) of the Department of our very variable among localities and even within a given Animal Biology (University of Salamanca, Salamanca, population. Dorsum always light or dark brown, greyish or Spain), CHUS01140319, holotype by present designation; even reddish, never green. Light dorsolateral stripes that adult male captured by Ana Pérez-Cembranos and Valentín can continue or not over parietal plates of the head. Supra- Pérez-Mellado on 14 September 2019 in the village of Galera dorsolateral stripes dark brown or black, present in almost (Granada province, Spain). all individuals. As in other Iberian Podarcis, light dorsolat- eral stripes are better defined in adult females. Upper half of flanks with a light brown spotted with small dark brown 4.2.2 | Type locality or black dashes, profusely reticulated with black spots or uniformly black, particularly in adult females. Lower half Galera, province of Granada (Spain). of flanks with a light brown sparsely spotted with brown or black small spots. Between upper and lower half of flanks, frequently there is a light brown stripe, again better 4.2.3 | Description of the holotype defined in adult females. Vertebral line is present in only a half of individuals (Table S2). When present, vertebral Adult male (Figure 1) with 48.5 mm of south–vent length and line is frequently bifurcated, especially on the upper half 115 mm of its intact tail length, 11.67 mm of pileus length, of dorsum. In a majority of adult individuals (Table S2), 5.62 mm of pileus width, 5.29 mm of head height, 15.85 mm belly is white or grey. However, a minor proportion of liz- of front leg length, 26.42 mm of hindleg length and 11.67 mm ards shows a yellowish or even an orange belly (Table S3 of foot length. Scalation: 56 longitudinal rows of dorsal and Figure S1). More than 40% of adult males have blue scales at mid-body (dorsalia), 25 gular scales (gularia), 10 ocelli on outer ventral scales. Masseteric plate is absent. collar scales (collaria), 21 transversal rows of ventral scales Juveniles, many adult females and even some adult males, from the collar to the anal plate (ventralia), 14 femoral pores can have a blue or greenish tail, sharply contrasted with on left hindleg (femoralia), 15 femoral pores on right hindleg, brown dorsum. 27 subdigital lamellae under the fourth toe, six submaxillary scales on each side, eight supralabial scales on each side, six supraocular scales on each side and 10 supraciliary granules 4.2.6 | Comparison with other species on each side. Coloration on the live animal: iris light brown. Pileus uniformly light brown. Dorsum brown, finely dotted. Table S3 shows the main differences between P. hispan- The individual lacks a well-formed vertebral line, although icus and P. galerai sp. nov., comparing coloration and there is an alignment of disconnected blackish points. Light design traits of both species with descriptions by Geniez brown on dorsolateral lines, well-marked and with sharp et al. (2007). Body size and general aspect of P. hispani- edges, with dark supra-dorsolateral bands formed by uncon- cus and P. galerai sp. nov. are quite similar. Males of nected black spots. Dorsolateral stripes continue over the P. galerai sp. nov. show a wider head width and a lower pileus. Tail greenish, clearly contrasted with the brown dor- number of femoral pores than P. hispanicus. Both sexes sum. Upper half of flanks dark brown, speckled with black. of P. galerai sp. nov. show a higher ventralia. In addi- Lower half of flanks with light brown background, stained tion, in P. galerai sp. nov. we observe a more pointed with dark brown. Ventral whitish-greyish. Gular area only snout and a more flattened head (Figure S1). There are with some black point in the lateral zones. Outer ventral several individuals that lack a vertebral line. When pre- scales with faded grey spots. Blue ocelli in all outer ventral sent, vertebral line is bifurcated in less than a half of scales. Submaxillary scales not pigmented (Figure S1). adult lizards (Table S3 and see above the diagnosis of P. galerai sp. nov.). In P. galerai sp. nov., belly is nor- mally white as in P. hispanicus, but there are individu- 4.2.4 | Etymology als with yellowish or even orange bellies. In addition, in several adult males it is possible to observe blue ocelli on The epithet galerai refers to the type locality: Galera, a vil- outer ventral scales, a trait only observed in a minority lage from north-eastern Granada province (Spain). of P. hispanicus males. BASSITTA et al. | 13 4.2.7 | Distribution and ecology 5 | Discussion

The geographical distribution of P. hispanicus and The mtDNA gene trees obtained show a clearer phylogeo- P. galerai sp. nov. is poorly known. According to our graphical pattern than in previous attempts to unravel the his- present-day data, P. hispanicus is approximately dis- tory of the species complex (see, for example, Kaliontzopoulou tributed from the latitude of Murcia town to the north. et al., 2011). The most basal node in the tree gives rise to four Apparently, P. hispanicus is today absent from its T. t. re- groups: a monophyletic group including P. muralis, a west- stricta, Monteagudo (Murcia province, Geniez et al., 2007 ern group (P. carbonelli, P. virescens, P. guadarramae and and pers. Obs.). We ignore the northern limit of this dis- P. bocagei), a south-eastern group (P. vaucheri, Valencia, tribution, especially in relation to P. liolepis. Similarly, P. hispanicus sensu stricto (Albacete/Murcia) and P. galerai the western limit remains to be clarified in relation to the sp. nov. (Galera)) and a low supported north-eastern group geographical area occupied by P. virescens. Our west- represented by P. liolepis (Figure 2). This low support could ernmost locality was Cañada del Provencio in Sierra be due to the lack of genetic data from the entire distribu- de Alcaraz (Albacete province). From our survey, the tion range of P. liolepis, as corroborated by the high genetic easternmost locality of P. hispanicus would be Callosa diversity showed on TCS networks for nuclear genes. An in- de Segura (Alicante province). Unfortunately, we can- teresting result was that P. liolepis was not found to be the not resolve if lizards from Elche (Alicante) can be as- sister group of P. galerai sp.nov., in contrast to what has been signed to P. hispanicus (Renoult et al., 2009). Podarcis indicated in other studies, which suggests that P. liolepis is galerai sp. nov. is mainly present to the south of Murcia a sister lineage to all remaining lineages of Iberian Podarcis town, apparently reaching coastal areas of Almería prov- (Kaliontzopoulou et al., 2011; Pinho et al., 2006). The species ince (between Dalias and Berja, Caeiro-Dias et al., 2018, delimitation and clustering analysis based on mtDNA identi- from data of Renoult et al., 2009). To the east, P. gale- fied three groups in the SE region (Valencia lineage, P. his- rai sp. nov. arrives to Embalse de la Pedrera, in Orihuela panicus sensu stricto and P. galerai sp. nov.). Even though (Alicante province). To the west, it would be present 2 km species delimitation analysis based on nuclear loci and com- to the north of Tíscar (Jaén province, Caeiro-Dias et al., bined data set (mtDNA + nuclear) provided support for treat- 2018). Caravaca de la Cruz (Murcia province) would be ment of the Valencia lineage, P. hispanicus sensu stricto and the northernmost known locality of P. galerai sp. nov. P. galerai sp. nov. as different species, the species tree output (Kaliontzopoulou et al., 2011). It is clear that we would in BPP and STACEY analyses could only provide signifi- need a deeper survey of geographical ranges of both spe- cant support for the early divergence of P. vaucheri, unlike cies. Particularly interesting will be to study overlap areas the findings based on mtDNA. All the species delimitation of both lizard species. According with known localities, analysis based on both mtDNA and nuclear loci reported the the altitudinal range of both species seems to be very same number of independent evolutionary units; therefore, similar (from 90 to 1,180 m.a.s.l. in P. hispanicus and no gene flow or introgression was detected, contrary to the 120 to 1,164 m in P. galerai sp. nov.). Both species are different number of clusters found using only mtDNA data, clearly saxicolous. As other Iberian species (i.e. P. gua- or morphological and nuclear data between P. hispanicus lin- darramae (Geniez et al., 2014 and references therein)), eages in Renoult et al. (2009). Networks based on nuclear P. hispanicus and P. galerai sp. nov. are mainly adapted genes indicated the uniqueness of P. muralis and P. vaucheri to rupicolous habitats. P. hispanicus is found in arid land- but showed a common haplotype between geographically scapes close to Murcia town (Laderas del Campillo), oc- close groups such as P. virescens, P. liolepis, P. galerai sp. cupying ruins and artificial walls with crevices. It is also nov., P. hispanicus sensu stricto or the Valencia lineage. A present in rocky outcrops inside pine forests (Cañada del posterior differentiation of the defined groups is also repre- Provencio). We also found lizards isolated in small ac- sented with the presence of species-specific haplotypes. cumulations of stacked stones that were removed from The SE region of the Iberian Peninsula presents a complex cultivated areas (Montealegre del Castillo). Podarcis geological history, highlighting the connection between the galerai sp. nov. occupies a range of arid Mediterranean Mediterranean Sea and the Atlantic Ocean through the Betic landscapes characterized by a strong human impact and corridor during the middle Miocene (Serravallian/Tortonian; a poor xerophytic and thermophilous vegetation. A vast 12.8–7.2 Ma) and the definitive closure during the Messinian proportion of the distribution range is a high plateau area (7.2–5.33 Ma) triggering the start of the Messinian Salinity characterized by a remarkable aridity and semi-desertic Crisis (5.97–5.33 Ma) connecting Africa and the Iberian conditions, where lizards are located in rocky outcrops Peninsula (Krijgsman et al., 1996, 2000, 2018). The geologi- on vacant lands (Puebla de Don Fadrique), in ruins of old cal instability present during these periods is thought to have buildings, artificial rock fences or in breakwaters of reser- caused the origin of many groups of organisms (Busack et al., voirs (Embalse de los Rodeos and Embalse de la Pedrera). 2005; Carranza et al., 2004; Kaliontzopoulou et al., 2011; 14 | BASSITTA et al. Pinho et al., 2006). We found that P. galerai sp. nov. split from naming of different lineages will allow these new species to the SE Iberian Podarcis group (Valencia lineage, P. hispanicus be considered in conservation policies (Geniez et al., 2007). sensu stricto and P. vaucheri) at 12.48 Ma. During this period, We can always expect new discoveries or deeper analyses to the Betic corridor was a connection between the Mediterranean change the general picture in the near future and, perhaps, cer- Sea and the Atlantic Ocean dividing the Iberian Peninsula at tain taxa will change status, such as the Valencia lineage. the Betic cordillera. This scenario could have caused the iso- Despite both phylogenetic and species delimitation anal- lation of P. galerai sp. nov. distributed in the south of this yses corroborated with multiple methodological approaches region. The separation between P. liolepis (distributed in the and using different data sets (mtDNA, nuclear loci or com- north-east) and the large clade of western lineages, including bined data) showed high supports for the different SE Podarcis P. carbonelli, P. virescens, P. guadarramae and P. bocagei, groups, genome-wide approaches would help to confirm these also occurred during this period (12.58 Ma). The divergence phylogenetic relationships and address the issues of introgres- between the Valencia lineage and the nominotypical taxon sion. Furthermore, to solve the puzzle of lineages of this species P. hispanicus sensu stricto occurred at 7.11 Ma. This period complex would be necessary to obtain a better picture of the corresponds with the end of the Tortonian, when the Betic distribution of Podarcis genus in the Iberian Peninsula. It could Corridor become colonizable by land with the definitive clo- be particularly important to study the P. liolepis and Valencia sure of the Betic Strait, causing the connection between the lineages in a deeper way, including a greater number of individ- north and south regions of the Iberian Peninsula and Africa. uals and locations. In addition, a global morphological study of According to our genetic results, lizards from Laderas del P. hispanicus complex would be of interest and would help to Campillo (Murcia province) belong to the Albacete/Murcia clarify the wide diversity presents in this species complex. lineage. This is the closest point to the restricted type locality In conclusion, this study provides a large genetic data set of of Monteagudo (Geniez et al., 2007) where we found Podarcis mtDNA and nuclear sequences of the P. hispanicus complex lizards. Even though P. galerai sp. nov. samples present mor- based on a greater number of individuals and locations than phologically similar traits, samples from the Albacete/Murcia in previous studies. These results allow a deeper phylogenetic lineage were identified as the nominal form of the complex, as analysis on SE Podarcis lineages that contributes to elucidat- this lineage shares almost all morphological features proposed ing the controversy regarding this region. Our phylogenetic by Geniez et al. (2007) for the name-bearing type specimens. tree shows the group including the Valencia lineage and P. his- In addition, our findings support a clear delimitation of panicus sensu stricto, as sister group of the North African form P. galerai sp. nov., on one hand, and the group that encom- (P. vaucheri). Podarcis galerai sp. nov. form a monophyletic passes the lineages from Valencia, P. hispanicus sensu stricto group, separated (~12 Ma) from the rest of the SE and North and North Africa, on the other. This separation is notewor- African forms, in contrast to previous studies based on a very thy and supports full species status for the Galera lineage. low number of individuals that showed the closest phyloge- The systematic status of the Valencia lineage remains unclear netic relationship between P. galerai sp. nov. and P liolepis with additional genetic information on the lineage, wider (Kaliontzopoulou et al., 2011; Pinho et al., 2006). Our results sampling of its geographical range and a full morphological support the elevation of P. galerai sp. nov. to full species rank. comparison with P. hispanicus sensu stricto and P. galerai sp. nov. required before full species status can be assigned. ACKNOWLEDGEMENTS It is difficult to stablish the sympatry or non-sympatry of We acknowledge Delegación Territorial de Granada de la different lineages/species of Podarcis on the Iberian Peninsula. Consejería de Agricultura, Ganadería, Pesca y Desarrollo Previous studies evidenced a parapatric distribution in the Sostenible de la Junta de Andalucía, for the licence for speci- eastern Iberian Peninsula; in contrast, in the west, species men captured in September 2019. like P. carbonelli and P. guadarramae are largely sympatric in their distribution ranges, as well as P. bocagei and P. gua- ORCID darramae. Further strict boundaries were proposed to separate Marta Bassitta https://orcid.org/0000-0001-6663-0820 eastern and southern species, such as P. vaucheri, P. liolepis, Ana Pérez-Cembranos https://orcid. P. virescens and P. hispanicus. The limits among these spe- org/0000-0002-8296-2283 cies of the complex are not yet well defined due to samples Valentín Pérez-Mellado https://orcid. being required from the entire distribution area. It is likely org/0000-0002-7137-4945 that the distribution areas will overlap, as it is the norm in Barbara Terrasa https://orcid.org/0000-0003-2607-4633 Western Iberia and in other areas of the Mediterranean basin. Richard P. Brown https://orcid.org/0000-0003-2401-6077 The only way to take into account the extraordinary diversity Jesús Ortega https://orcid.org/0000-0001-6599-1675 of mitochondrial lineages of Iberian Podarcis is to document Jose A. Castro https://orcid.org/0000-0001-8262-560X their morphological characteristics and make taxonomic deci- Antònia Picornell https://orcid.org/0000-0003-4454-5304 sions. We agree with the important argument that only formal Cori Ramon https://orcid.org/0000-0003-1234-3337 BASSITTA et al. | 15 REFERENCES Sceloporus scalaris species group (Squamata: Phrynosomatidae). Systematic Biology 63 Arnold, E. N., Arribas, O., & Carranza, S. (2007). Systematics of , (2), 119–133. the Palaearctic and Oriental lizard tribe Lacertini (Squamata: Harris, D. J., & Sa-Sousa, P. (2002). Molecular phylogenetics of Iberian Podarcis Podarcis hispanica Lacertidae: Lacertinae), with descriptions of eight new genera. wall lizards ( ): is a species complex? Molecular Phylogenetics and Evolution 23 Zootaxa, 1430(1), 1–86. , (1), 75–81. Bouckaert, R., Vaughan, T. G., Barido-Sottani, J., Duchêne, S., Heled, J., & Drummond, A. J. (2010). Bayesian inference of species Molecular Biology and Evolution 27 Fourment, M., Gavryushkina, A., & Matschiner, M. (2019). BEAST trees from multilocus data. , , 2.5: An advanced software platform for Bayesian evolutionary anal- 570–580. ysis. PLoS Computational Biology, 15(4), e1006650. Hoang, D. T., Chernomor, O., Von Haeseler, A., Minh, B. Q., & Vinh, Brown, R. P., Terrasa, B., Pérez-Mellado, V., Castro, J. A., Hoskisson, L. S. (2017). UFBoot2: improving the ultrafast bootstrap approxi- Molecular Biology and Evolution 35 P. A., Picornell, A., & Ramon, M. M. (2008). Bayesian estima- mation. , (2), 518–522. tion of post-Messinian divergence times in Balearic Island lizards. Jones, G. (2017). Algorithmic improvements to species delimitation and Journal of Molecular Phylogenetics and Evolution, 48(1), 350–358. phylogeny estimation under the multispecies coalescent. Mathematical Biology 74 Brown, R. P., & Yang, Z. (2011). Rate variation and estimation of di- , , 447–467. vergence times using strict and relaxed clocks. BMC Evolutionary Kaliontzopoulou, A., Carretero, M. A., & Llorente, G. A. (2012). Podarcis Biology, 11(1), 271. Morphology of the wall lizards (Squamata: Lacertidae) Buades, J. M., Rodríguez, V., Terrasa, B., Pérez-Mellado, V., Brown, from the Iberian Peninsula and North Africa: patterns of variation Zoological Journal of the R. P., Castro, J. A., … Ramon, M. M. (2013). Variability of the in a putative cryptic species complex. Linnean Society 164 mc1r gene in melanic and non-melanic Podarcis lilfordi and , (1), 173–193. Podarcis pityusensis from the Balearic Archipelago. PLoS One, Kaliontzopoulou, A., Pinho, C., Harris, D. J., & Carretero, M. A. 8(1), e53088. (2011). When cryptic diversity blurs the picture: a cautionary tale Podarcis Biological Busack, S. D., Lawson, R., & Arjo, W. M. (2005). Mitochondrial DNA, from Iberian and North African wall lizards. Journal of the Linnean Society 103 allozymes, morphology and historical biogeography in the Podarcis , (4), 779–800. vaucheri (Lacertidae) species complex. Amphibia-Reptilia, 26, Kapli, P., Lutteropp, S., Zhang, J., Kobert, K., Pavlidis, P., Stamatakis, 239–256. A., & Flouri, T. (2017). Multi-rate poisson tree processes for sin- Caeiro-Dias, G., Luís, C., Pinho, C., Crochet, P. A., Sillero, N., & gle-locus species delimitation under maximum likelihood and Bioinformatics 33 Kaliontzopoulou, A. (2018). Lack of congruence of genetic and Markov chain Monte Carlo. , , 1630–1638. niche divergence in Podarcis hispanicus complex. Journal of Katoh, K., & Toh, H. (2008). Recent developments in the MAFFT mul- Briefings in Bioinformatics 9 Zoological Systematics and Evolutionary Research, 56(4), 479–492. tiple sequence alignment program. , (4), Carranza, S., Arnold, E. N., & Amat, F. (2004). DNA phylogeny 286–298. Color catalogue for field biologists of Lacerta (Iberolacerta) and other lacertine lizards (Reptilia: Köhler, G. (2012). . Offenbach, Lacertidae): did competition cause long-term mountain restriction? Germany: Herpeton Verlag. Systematics and Biodiversity, 2(1), 57–77. Krijgsman, W., Capella, W., Simon, D., Hilgen, F. J., Kouwenhoven, Clement, M., Posada, D., & Crandall, K. A. (2000). TCS: a computer T. J., Meijer, P. T., … Flecker, R. (2018). The Gibraltar corridor: Marine Geology 403 program to estimate gene genealogies. Molecular Ecology, 9, Watergate of the Messinian salinity crisis. , , 1657–1659. 238–246. Corander, J., & Tang, J. (2007). Bayesian analysis of population structure Krijgsman, W., Garcés, M., Agustí, J., Raffi, I., Taberner, C., & based on linked molecular information. Mathematical Biosciences, Zachariasse, W. J. (2000). The ‘Tortonian salinity crisis’ of the east- Earth and Planetary Science Letters 181 205, 19–31. ern Betics (Spain). , (4), Fujisawa, T., & Barraclough, T. G. (2013). Delimiting species using 497–511. single-locus data and the Generalized Mixed Yule Coalescent ap- Krijgsman, W., Garcés, M., Langereis, C. G., Daams, R., Van Dam, J., proach: a revised method and evaluation on simulated data sets. Van der Meulen, A. J., … Cabrera, L. (1996). A new chronology for Earth and Systematic Biology, 62(5), 707–724. the middle to late Miocene continental record in Spain. Planetary Science Letters 142 Geniez, P., Cluchier, A., Sá-Sousa, P., Guillaume, C. P., & Crochet, , (3–4), 367–380. P. A. (2007). Systematics of the Podarcis hispanicus-complex Kumar, S., Stecher, G., & Tamura, K. (2016). MEGA7: molecular evo- Molecular (Sauria, Lacertidae) I: Redefinition, morphology and distribution lutionary genetics analysis version 7.0 for bigger datasets. Biology and Evolution 33 of the nominotypical taxon. The Herpetological Journal, 17(2), , (7), 1870–1874. 69–80. Lanfear, R., Frandsen, P. B., Wright, A. M., Senfeld, T., & Calcott, B. Geniez, P., Sa-Sousa, P., Guillaume, C. P., Cluchier, A., & Crochet, P. (2016). PartitionFinder 2: new methods for selecting partitioned A. (2014). Systematics of the Podarcis hispanicus complex (Sauria, models of evolution for molecular and morphological phylogenetic Molecular Biology and Evolution 34 Lacertidae) III: valid nomina of the western and central Iberian analyses. , (3), 772–773. forms. Zootaxa, 3794(1), 1–51. Leaché, A. D., Zhu, T., Rannala, B., & Yang, Z. (2018). The spectre of Systematic Biology 68 González, P., Pinto, F., Nogales, M., Jiménez-Asensio, J., Hernández, M., too many species. , (1), 168–181. & Cabrera, V. M. (1996). Phylogenetic relationships of the canary Leigh, J. W., & Bryant, D. (2015). PopART: Full-feature software for Methods in Ecology and Evolution islands endemic lizard Genus Gallotia (Sauria: Lacertidae), inferred haplotype network construction. , 6 from mitochondrial DNA sequences. Molecular Phylogenetics and , 1110–1116. Evolution, 6(1), 63–71. Librado, P., & Rozas, J. (2009). DnaSP v5: a software for comprehen- Bioinformatics 25 Grummer, J. A., Bryson, R. W. Jr, & Reeder, T. W. (2014). Species de- sive analysis of DNA polymorphism data. , (11), limitation using Bayes factors: simulations and application to the 1451–1452. 16 | BASSITTA et al. Minh, B. Q., Nguyen, M. A. T., & von Haeseler, A. (2013). Ultrafast Lacertidae) II: the valid name of the north-eastern Spanish form. approximation for phylogenetic bootstrap. Molecular Biology and Zootaxa, 2500(1), 58–68. Evolution, 30(5), 1188–1195. Rodríguez, V., Brown, R. P., Terrasa, B., Pérez-Mellado, V., Castro, J. Nguyen, L. T., Schmidt, H. A., von Haeseler, A., & Minh, B. Q. A., Picornell, A., & Ramon, M. M. (2013). Multilocus genetic di- (2014). IQ-TREE: a fast and effective stochastic algorithm for es- versity and historical biogeography of the endemic wall lizard from timating maximum-likelihood phylogenies. Molecular Biology and Ibiza and Formentera, Podarcis pityusensis (Squamata: Lacertidae). Evolution, 32(1), 268–274. Molecular Ecology, 22(19), 4829–4841. Pérez-Mellado, V., & Gosá, A. (1988). Biometría y Folidosis en Rodríguez, V., Brown, R. P., Terrasa, B., Pérez-Mellado, V., Picornell, Lacertidae (Sauria, Reptilia). Algunos aspectos metodológicos. A., Castro, J. A., & Ramon, C. (2014). Genetic diversity and his- Revista Española De Herpetología, 3(1), 97–104. torical biogeography of the Maltese wall lizard, Podarcis filfo- Pinho, C., Ferrand, N., & Harris, D. J. (2006). Reexamination of the lensis (Squamata: Lacertidae). Conservation Genetics, 15(2), Iberian and North African Podarcis (Squamata: Lacertidae) phylog- 295–304. eny based on increased mitochondrial DNA sequencing. Molecular Rodríguez, V., Buades, J. M., Brown, R. P., Terrasa, B., Pérez-Mellado, Phylogenetics and Evolution, 38(1), 266–273. V., Corti, C., … Ramon, M. M. (2017). Evolutionary history of Pinho, C., Harris, D. J., & Ferrand, N. (2007). Comparing patterns of nu- Podarcis tiliguerta on Corsica and Sardinia. BMC Evolutionary clear and mitochondrial divergence in a cryptic species complex: the Biology, 17(1), 27. case of Iberian and North African wall lizards (Podarcis, Lacertidae). Ronquist, F., Teslenko, M., Van Der Mark, P., Ayres, D. L., Darling, Biological Journal of the Linnean Society, 91(1), 121–133. A., Höhna, S., … Huelsenbeck, J. P. (2012). MrBayes 3.2: efficient Pinho, C., Harris, D. J., & Ferrand, N. (2008). Non-equilibrium esti- Bayesian phylogenetic inference and model choice across a large mates of gene flow inferred from nuclear genealogies suggest that model space. Systematic Biology, 61(3), 539–542. Iberian and North African wall lizards (Podarcis spp.) are an as- Stephens, M., Smith, N. J., & Donnelly, P. (2001). A new statistical semblage of incipient species. BMC Evolutionary Biology, 8(1), 63. method for haplotype reconstruction from population data. The Pons, J., Barraclough, T. G., Gomez-Zurita, J., Cardoso, A., Duran, D. American Journal of Human Genetics, 68(4), 978–989. P., Hazell, S., … Vogler, A. P. (2006). Sequence-based species de- Terrasa, B., Pérez-Mellado, V., Brown, R. P., Picornell, A., Castro, J. limitation for the DNA taxonomy of undescribed . Systematic A., & Ramon, M. M. (2009). Foundations for conservation of in- Biology, 55(4), 595–609. traspecific genetic diversity revealed by analysis of phylogeograph- Puillandre, N., Lambert, A., Brouillet, S., & Achaz, G. (2012). ABGD, ical structure in the endangered endemic lizard Podarcis lilfordi. Automatic Barcode Gap Discovery for primary species delimitation. Diversity and Distributions, 15(2), 207–221. Molecular Ecology, 21(8), 1864–1877. Yang, Z. (1993). Maximum likelihood estimation of phylogeny from R Core Team (2018). R: A language and environment for statistical DNA sequences when substitution rates differ over sites. Molecular computing. Vienna, AustriaR Foundation for Statistical Computing. Biology and Evolution, 10, 1396–1401. Retrieved from https://www.R-proje​ct.org/ Yang, Z., & Rannala, B. (2010). Bayesian species delimitation using Rambaut, A. (2014). FigTree 1.4.2 software, Institute of Evolutionary multilocus sequence data. Proceedings of the National Academy of Biology. Edinburgh, UKUniversity of Edinburgh. Sciences of the United States of America, 107(20), 9264–9269. Rambaut, A., Drummond, A. J., Xie, D., Baele, G., & Suchard, M. A. Zhang, J., Kapli, P., Pavlidis, P., & Stamatakis, A. (2013). A general spe- (2018). Posterior summarization in Bayesian phylogenetics using cies delimitation method with applications to phylogenetic place- Tracer 1.7. Systematic Biology, 67(5), 901–904. ments. Bioinformatics, 29(22), 2869–2876. Rannala, B. (2015). The art and science of species delimitation. Current Zoology, 61(5), 846–853. Rannala, B., & Yang, Z. (2003). Bayes estimation of species divergence SUPPORTING INFORMATION times and ancestral population sizes using DNA sequences from Additional supporting information may be found online in multiple loci. Genetics, 164(4), 1645–1656. the Supporting Information section. Rannala, B., & Yang, Z. (2017). Efficient Bayesian species tree infer- ence under the multispecies coalescent. Systematic Biology, 66(5), 823–842. How to cite this article: Bassitta M, Buades JM, Renoult, J. P., Geniez, P., Bacquet, P., Benoit, L., & Crochet, P. A. Pérez-Cembranos A, et al. Multilocus and (2009). Morphology and nuclear markers reveal extensive mito- morphological analysis of south-eastern Iberian Wall chondrial introgressions in the Iberian Wall Lizard species complex. lizards (Squamata, Podarcis). Zool Scr. 2020;00:1–16. Molecular Ecology, 18(20), 4298–4315. https://doi.org/10.1111/zsc.12450 Renoult, J. P., Geniez, P., Bacquet, P., Guillaume, C. P., & Crochet, P. A. (2010). Systematics of the Podarcis hispanicus-complex (Sauria,